Feature Extraction And Classification For Images

With the development of science and technology, it is possible for us to obtain large numbers of images from image equipment, networks and video databases. As a result, the request for image processing is becoming higher and higher. This greatly boosts the development of image processing technology. The main research contents for image processing include image enhancement, image recovery, image code and image classification. This paper mainly deals with feature extraction and classification for images.The quality of extracted features is the key factor to decide the classification performance. Although there are many ways to describe image, texture is still a basic feature for image analysis which is hard to describe. A method for image classification based on feature fusion is put forward which combines the characteristics of texture images with frame wavelet transform, considers the relations among scales and textural information included in different frequency domains. With respect to large class number and high-dimension feature vector, support vector machine is used as classifier. The experimental results indicate that the method above has enhanced computational speed and recognition rate of classes.By studying image features and classification, a method of object detection based on feature extraction and mathematical morphology is proposed in this paper. For objects in heavily cluttered environments, a preprocessing step is employed in order to reduce the effect of background. Compared with the traditional global segmentation, the proposed method is more efficient and simple which yields the same accurate detection results even if the background is more complex.Curvelet transform is able to represent smooth and edge parts of image with sparsity which can provide more image information than wavelet transform. An idea of feature extraction based on curvelet transform is presented. The experimental results prove that the method can perfectly represent the texture directionality.